Speaker: | Laszlo Ladanyi |
Thomas J. Watson Research Center, IBM |
Joint work with Marta Eso and David Jensen.
Title: Solving Lexicographic Multiobjective MIPs with Branch-Cut-Price
Presentation Slides
In this talk we will first describe an application, the FCC Auction # 31, and its two formulation: one with a compact representation solvable via branch-and-cut, and one with column generation using branch-and-price. Then we will show how to utilize Dantzig-Wolfe decomposition to perform branch-cut-price on general MIPs and illustrate how this applies to the column generation version of our application. Finally, we extend the applicability of linear complementarity to integer programming to solve problems with multiple (lexicographic) objective.